Robust Learning

Robust learning aims to develop machine learning models that are resilient to various forms of noise and uncertainty in data, including label noise, adversarial attacks, and distribution shifts. Current research focuses on developing algorithms and model architectures (e.g., multiview SVMs, graph neural networks, and diffusion models) that incorporate techniques like adversarial training, data augmentation, and loss function modifications to enhance robustness. These advancements are crucial for improving the reliability and generalizability of machine learning models in real-world applications, particularly in safety-critical domains like healthcare and autonomous systems, where data imperfections are common.